Safe nonlinear controller switching with biomedical engineering applications

Abstract

Suppose an unknown or only partially identified plant is operating with a known linear controller in a closed-loop, and the use of a known nonlinear controller appears attractive to replace the existing controller, on the basis of limited knowledge of the plant. We apply our novel databased verification tests for ensuring that the introduction of the new nonlinear controller will stabilize the true plant for two different practical scenarios. First, we explore a dynamic mass system to illustrate how our tests can be utilized for verifying a nonlinear regulator given as a state feedback law. Second, we apply our tests to a biomedical engineering control problem in a multiple model adaptive control setting, where a nonlinear blood pressure regulator is required for better performance. Show more